GeoAI-Driven Customised Geocoding System for University Campus Leveraging Geohashing, VGI and NLP Techniques: A Case Study of IIT Kharagpur

Location-based services (LBS) are essential to modern urban infrastructure, enabling seamless navigation and spatial awareness. However, their effectiveness depends on accurate addressing systems, which traditional geocoding methods often fall short, especially in complex environments like university campuses. These settings feature a mix of formal addresses, unaddressed spaces, informal landmarks, and constantly evolving infrastructure, making navigation and management challenging. Since conventional geocoding relies on rigid, structured address systems, it lacks the adaptability and granularity needed for dynamic campus environments. In this context, the study introduces a novel GeoAI-driven customized geocoding system specifically designed for university campuses. The system primarily integrates three key technologies: Geohashing for efficient spatial indexing, Volunteered Geographic Information (VGI) for enriched and dynamic location data, and Natural Language Processing (NLP) for intelligent, user-friendly location retrieval. Customized, human-readable geocodes form the foundation of this approach, enabling precise and intuitive location referencing. These geocodes are further integrated into a campus specific digital gazetteer, which leverages Volunteered Geographic Information (VGI) to capture both formal locations and evolving informal landmarks. Additionally, the geocoding system supports an NLP-driven query mechanism, enabling accurate interpretation of user queries and intuitive campus-level navigation. The findings demonstrate strong performance across both NLP components. The Named Entity Recognition (NER) model achieves a precision of 0.887, a recall of 0.853, and an F1-score of 0.870 in identifying campus-specific spatial entities, while the BERT-based model attains an F1-score of 0.82 in resolving complex location-based queries. To ensure seamless geospatial exploration, the proposed system is deployed as a WebGIS application with voice-enabled search and navigation, providing an interactive and efficient interface for accessing location information. The proposed framework advances geocoding and location- based services (LBS) by enabling scalable, adaptable, and context-aware addressing, with potential applicability to complex spatial environments beyond university campuses.